1. Introduction
Pre-Tertiary fractured basement (referred to below as “fractured basement”) is an important type of hydrocarbon-bearing reservoirs off shore Vietnam. In the last decade, thanks to advanced seismic technology, seismic imaging of fracture zones associated with faults within the basement has signifi cantly improved. Seismic attributes can, therefore, be widely applied to identify and outline good fracture zones inside the basement as well as to predict some of the main characteristics of existing fracture systems, such as dip and azimuth [2, 4].
Currently, the application of seismic attributes and well data for fractured basement studies in Vietnam could be classifi ed in two groups. Based on detailed analysis of diff erent seismic attributes the authors of the fi rst group [1 - 4] have proved the possibility of using seismic attributes in the identifi cation and outlining of good fracture zones inside the basement as well as in the prediction of fracture systems’ characters. The authors of the second group [5] have tried to build geological model of the fractured basement based on integrated seismic attributes and well data using ANN and Co-Kriging technique without detailed analysis of the applied seismic attributes.
This paper demonstrates the good results of combining the ideas of both author groups in the construction of a porosity model for the Hai Su Den fractured basement reservoir. The workfl ow of porosity model building for
the Hai Su Den basement is consistent with two steps. In the fi rst step, meaningfull seismic attributes have been selected based on a detailed analysis of the possibility for basement fracture imaging and characterisation.
The second step builds a porosity model by integrating the ANN and Co-Kriging technique. The fi nal model has a good correlation with blind test well results and a high degree of reliability.
2. Hai Su Den fi eld overview and study method 2.1. Hai Su Den fi eld overview
The Hai Su Den oil fi eld is located in Block 15-2/01 of Cuu Long basin, off shore the South of Vietnam (Fig.1). The Hai Su Den structure is an anticlinal drape over a basement high, which is elongated along the NE-SW direction. At the Early Miocene, Oligocene and basement levels the structure is intersected by a series of E-W to NW-SE and NE-SW faults (Figs.2 & 3).
Previous studies in the area suggested that the granitoid basement in Block 15-02/01 is considered to be Cretaceous Deo Ca suite granitoid. It is possibly constituted of the Cretaceous-Paleogene mafi c, intermediate and felsic dykes (Deo Ca, Cu Mong, Phan Rang suites). The basement could be strongly hydrothermal altered.
Secondary minerals were formed and fi lled fractures and/
or partly replaced primary minerals. The most common alterations are sericitisation, calcitisation of plagioclase, kaolinisation of potassium feldspar, chloritisation of
BUILDING POROSITY MODEL OF FRACTURED BASEMENT RESERVOIR BASED ON INTEGRATED SEISMIC AND WELL DATA IN HAI SU DEN FIELD, CUU LONG BASIN
Nguyen Anh Duc1, Nguyen Huy Ngoc2, Nguyen Lam Anh3
1Petrovietnam Exploration Production Corporation (PVEP), 2Petronas Carigali Sdn Bhd (PCSB), 3Research and Design Institute - Vietsovpetro
The Pre-Tertiary fractured basement is an important hydrocarbon-bearing reservoir in Vietnam. Due to their very small matrix porosity, basement rocks become reservoirs only when they are strongly fractured, consequently it is a big challenge to construct the porosity model for the basement. Based on the good seismic imaging of the Hai Su Den basement, diff erent seismic attributes have proved to be eff ective tools in basement fracture characterisation, and the integration of 3D seismic attributes, well data and other geological information by using Artifi cial Neural Network (ANN) and Co-Kriging techniques has been demonstrated as a good way to construct porosity model for the Hai Su Den fractured basement reservoir. The accuracy of the model has been verifi ed by well results.
Key words: Fractured basement, Pre-Tertiary, porosity, permeability, Hai Su Den fi eld.
Summary
biotite and hornblende. There are some veins of zeolite observed on core [8].
Seven wells have been drilled into the Hai Su Den basement, which are HSD-1X, HSD-2X/ST, HSD-3X, HSD-4X, HSD-5XP, VD-1X and VD-2X. HSD-1X drilled in Block B in September 2007 has a maximum fl ow rate of more than 20,000 bop while VD-1X, which was drilled before in the same basement block and is only about 600 metres away from HSD-1X surface location, was dry.
These results show very complicated reservoir characters of the Hai Su Den fractured basement.
Source rock consists of mainly shales of the Oligocene D sequence with high TOC and HI values, which are considered good to excellent oil-prone source rocks. The lacustrine shales of Early Oligocene E sequence are also considered as having good source rock potential.
Trap: The fractured basement reservoir is known to be formed in the Hai Su Den prospect from existing well control. Multiple trending faults are present within the neighbourhood of the structure. This fault pattern suggests the presence of a complex of shear and extensional fractures.
Seal: Thick shales of the D Sequence are main seals for the clastic reservoirs of the E Sequence and/or basement reservoirs. Intra-formational shales, interbedded within sandstones in the E, C and BI sequences may also be potential seals, or have partially sealing capacity for the immediately underlying sandstone reservoirs.
Reservoir: The fractured/weathered granite basement is the primary reservoir objective in the Cuu Long basin, from which hydrocarbon has been produced from fractured zones within the pre-Tertiary basement. The secondary reservoir objectives are the F ig.1. Hai Su Den oil fi eld location map
Fig.2. Stratigraphy and main deformation phases in Block 15-02/01, Cuu Long basin
Fig.3. Depth structural map of Hai Su Den basement top with results of FMI interpretation.
The azimuths of fractures interpreted by FMI data are consistent with fault direction interpreted by seismic data
reservoirs within the Oligocene fl uvio-deltaic sandstones and the Lower Miocene sandstones, from which oil has been discovered.
In the Hai Su Den basement, three fault systems with E-W, NE-SW and NW-SE directions are strongly developed and fractures associated with these fault systems interpreted by FMI data are consistent with the interpreted faults in their characters (Fig.3). Within the Hai Su Den basement, the role of the E-W fault system was also highlighted by the previous studies [2] as the main faults related with oil fl ow zones inside the basement.
2.2. Study method
Several methods have been applied to build the porosity model for fractured basement in Cuu Long basin including Halo model/Advanced Halo model, DFN (Discrete Fracture Network) and ANN [5].
The Halo model has been widely applied in Vietnam, which much depends on the fault interpretation and assumes that the porosity remains unchanged along the interpreted fault planes. In the Hai Su Den basement, the Halo model method has been used at the beginning, but the results were too optimistic, therefore the advanced Halo model method has been tested. The diff erence between the Halo model and advanced Halo model methods is mainly related to applying seismic attributes to take into account the discontinuity of reservoir quality along the interpreted fault planes.
Besides, DFN is also a method which can create the connectivity between faults and fracture systems from the geological information of an area. This method requires several information of a fracture system including orientation, aperture, intensity, location, mineralogy, hydraulic and mechanical properties [9]. However, DFN has the same disadvantage as the Halo model since it does not refl ect the heterogeneity of fracture systems. In order to promote the role of seismic attributes in porosity model building process, ANN has been applied [5]. This method is able to combine seismic and Fig.4. Comparison of diff erent seismic data in Hai Su Den basement [4]
Fig.6. Rel ative acoustic impedance (RAI) gives much better images inside basement elements than the original seismic data
Fig.5. Good tie of dipping refl ection events inside the Hai Su Den basement with synthetic seismogram (red fi lled curve) and fracture porosity (Green curve) derived
from well log data by using BASROCK software
well data to produce a porosity model which refl ects the vertical and horizontal porosity variation trends. The main weakness of the current method applying ANN in Cuu Long basin is the lack of a detailed analysis of the applied seismic attributes in imaging and characterisation of fracture systems.
The quality of 3D seismic data of the Hai Su Den basement is good and could clearly image fracture associated faults. Fig.4 shows a comparison between 4 seismic data versions of the Hai Su Den area (2006 PSTM, 2007 Kirchoff PSDM, 2007 Beam PSDM and 2009 Beam PSDM), which were the result of applying diff erent processing sequences. It is clear that the 2009 Beam PSDM data has better quality while it could clearly image dipping elements inside the basement. Beam migration can enhance the signal to noise ratio,
especially enhance steeply dipping events, handle multi-arrival ray paths and preserve steep dipping refl ection, hence providing a clear image of the basement [1]. By using synthetic seismograms, it’s clearly to see that the dipping refl ections inside the basement have good tie with fracture zones generated from logging data.
The synthetic seismogram in Fig.5 shows good consistency between seismic and well data.
To avoid the above mentioned weakness of current application of the ANN method, under the condition that good seismic imaging of fracture systems was achieved in the Hai Su Den fractured basement, an integration of ANN and Co-Krigi ng techniques was used for porosity model building for Hai Su Den fractured basement reservoir with the following workfl ow:
- Detailed analysis of seismic attributes, which could image and characterise fracture systems inside the basement to select meaningful fracture porosity and the optimum seismic attribute set.
- Porosi ty model building using ANN with the selected seismic
attribute set and fracture porosity estimated by BASROCK software.
- Porosity model building using Co-Kriging to inte- grate seismic attributes, fracture porosity and other well data and fracture characters collected in the region.
3. Possi bility of using seismic attributes to build po- rosity model of the fractured basement reservoir in Hai Su Den fi eld
Recently, many seismic attributes have been successfully applied to image and characterise the fracture systems inside the basement. Among them, the following seismic attributes have been highlighted in diff erent publications from the previous studies [1 - 4, 7]: Relative acoustic impedance (RAI), Variance (Coherency), Curvatures,
Fig.7. Verti cal section of diff erent seismic attributes showing diff erent quality in imaging of fracture zones inside the Hai Su Den basement. A: Variance; B: Ant-Tracking;
C: Cosine of Phase; D: Envelope
Fig.8. Vertical section of diff erent seismic attributes showing diff erent quality in imaging of fracture zones inside the Hai Su Den basement. A: RMS Amplitude;
B: Gradient magnitude; C: Sweetness; D: Refl ection intensity A
C D
B
RMS amplitude, Envelope , Ant-trac king, Cosine o f phase, Gradient magnitude, Sweetness and refl ection intensity.
Based on our study results, the relative acoustic impedance is the most important attribute. At fi rst, the seismic inversion process normally increases the signal to noise ratio, consequently it improves dipping refl ection events inside the basement. Fig.6 shows this advantage in the Hai Su Den basement. Secondly, RAI is a layer property, which is diff erent from the surface property of the original seismic data, it, therefore, directly refl ects the fracture zones inside the basement, which are characterised by lower density and lower seismic velocity (Lower RAI) compared to the fresh basement. For the Hai Su Den basement the RAI was selected as the input data instead of the original seismic data to assess other seismic attributes.
To select the meaningful fracture porosity and optimum seismic attribute sets, the seismic attribute analysis is performed in two steps:
- Qualitative analysis to select the meaningful fracture porosity: The qualitative analysis is based on visual inspection of seismic attributes in im- aging of fracture systems and qualita- tive correlation between seismic attri- butes and fracture porosity.
- Quantitative analysis to select optimum seismic attribute set: The quantitative analysis calculates and ranks correlation coeffi cients between seismic attributes and well fracture po- rosity. The optimum seismic attribute set is selected by the highest correla- tion coeffi cient among diff erent com- bination of the seismic attributes.
3.1. Quali tative analysis of seismic attributes
3.1.1. Variance and curvature
These attributes are well known in fault and fracture imaging for sediment section and top of the basement [1, 3, 7], but for inside the basement these attributes become too noisy to image the fracture systems. Fig.7 presents the
vertical section of the variance attribute for inside the Hai Su Den basement. It is clear that the attribute anomalies exist everywhere and many of them are vertical dipping events; they are, therefore, probably related to noise.
These attributes were not selected for further quantitative analysis.
3.1.2. Ant-Tracking
Ant-tracking is a well known attribute for fault mapping and fracture system characterisation for both sediment and inside-basement sections [3]. Based on this attribute we can successfully predict azimuth, dip angle, density and intersection between diff erent fracture systems (Fig.7B). But it is diffi cult to identify and outline high fractured zones inside the basement using this attribute. In principle the Ant-tracking will gain the weak
Fig.10. Comparison between gradient magnitude and HSD-5XP well fracture porosity (red curve) on both section and depth slices
Fig.9. Comparison betw een sweetness and HSD-5XP wel l fracture porosity (red curve) on both section and depth slices
seismic events, therefore it does not preserve the true amplitude and reduce discontinuity along fault plane, which is an unexpected eff ect in fracture porosity model building. In Fig.7B we can see good fault imaging, but the anomaly intensity is similar everywhere in the section and it is hard to be correlated with variation of fracture porosity with depth.
3.1.3. Cosine of phase
Similar to the Ant-Tracking attribute, Cosine of instantaneous phase could be applied for inside fault imaging and fracture systems characterisation (Fig.7C), but it is hard to be correlated with fracture porosity.
3.1.4. Envelope
Envelope or instantaneous amplitude is the complex seismic trace, which is related to the intensity of AI contrast between fractured and fresh basement and thickness of the fracture zones, it could, therefore, not only image good fracture zones inside the basement but also be used for fracture system characterisation. Fig.7D presents the section of the Envelope attribute for the Hai Su Den basement. In this fi gure we can see that strong Envelope anomalies exist in diff erent areas and reduce with depths.
The strong envelope anomaly zones are also consistent with location of good oil fl ow basement wells. This attribute was defi nitely selected for quantitative analysis.
3.1.5. RMS Amplitude
RMS Amplitude computes Root Mean Squares on
instantaneous trace samples over a specifi ed window.
Similar to the Envelope attribute it could be used for both identifi cation and outline of good fracture zones inside the basement as well as fracture system characterisation.
Fig.8A presents RMS amplitude section in the Hai Su Den Basement.
3.1.6. Gradien t magnitude
The magnitude of the instantaneous gradient computed in three-dimensions of the sample neighbourhood. This attribute can highlight the lineament of faults and indicate zones of high-fractured density, thus enhance the signature of fractured zones inside the basement (Fig.8B).
3.1.7. Sweetne ss
The sweetness attribute is calculated using the following formula: “Sweetness = Env/sqrt (Inst. Freq)”, it refl ects both the contrast between the fractured and fresh basement and the fracture zones thickness itself. Within a fractured zone, the frequency is also reduced due to attenuation of energy. Fig.8C is the sweetness section for the Hai Su Den basement.
3.1.8. Refl ect ion Intensity
Refl ection intensity is the average amplitude over a specifi ed window multiplied with the sample interval.
Visually, refl ection intensity can help to highlight the lineament of faults and indicate zones of high fracture density (Fig.8D).
3.2. Quantitative analysis of the selected seismic attributes
The latest six seismic attributes and RAI were chosen for quantitative analysis to select the optimum seismic attributes for further porosity model building using ANN.
Based on the quantitative analysis, the three following seismic attributes were selected as the optimum fracture porosity attributes.
3.2.1. Sweetness
Fig.9 shows a comparison between the sweetness attribute and well fracture porosity in both Fig.11. Comparison between refl ection intensity and HSD-5XP well fracture porosity
(red curve) on both section and depth slices
vertical section and diff erent depth slices. It is clear to see that the seismic attribute (blue colour) is consistent with the well data not only in location of good fracture zones, but also with dip and azimuth of the fracture systems.
3.2.2. Gradient magnitude
Gradient magnitude attribute has a good tie with the well fracture porosity in location of good fracture zones and in dip and azimuth of diff erent fracture systems (Fig.10).
3.2.3. Refl ection Intensity
Similar to the Sweetness and Gradient magnitude attributes, Refl ection Intensity also has a very good tie with the well data (Fig.11).
4. Porosity model building using ANN and Co-Kriging techni que 4.1. ANN
ANN is applied to integrate the selected seismic attributes with well fracture porosity to predict the distribution of porosity within basement.
ANN is a tool for automatically fi nding the relationship between multiple known par ameters and a single unknown parameter. The behaviour of a neural network is defi ned by the way its individual
computing elements are connected and by the strength of those connections, or weights. The weights are automatically adjusted by training the network according to a specifi ed learning rule until it properly performs a desired task.
Supervised ANN is an algorit hm that takes multiple inputs and returns one or several outputs. These inputs may match with log values, seismic attributes, surface values or properties from the same cell. Each input is multiplied by a weight; the result is summed and passed through a nonlinear function to produce an output [5].
The set of 03 selected seismic attributes (Reflection Intensity, Gradient Magnitude, and Sweetness) was run in supervised mode using ANN and well fracture porosity was used as training data. We can observe that this optimum seismic attribute set has a good correlation coeffi cient (as 0.76) with well porosity, therefore the result of predicted porosity by ANN is highly reliable.
Figs.12 and 13 show a very good consistence of well fracture porosity with the fracture porosity predicted by using ANN.
Fig.13. Porosity model and depth slices using ANN of Hai Su Den fractured basement Fig.12. Vertical section of porosity model using ANN.
The red curves are well fracture porosity
4.2. Co-Kriging technique
Based on the result of predicted fracture porosity model using ANN, Co-Kriging steps have been taken to correct the trend and extent of fractures in the model by well parameters, fracture parameters and tectonic elements of the study area. In this process, the model was used as a secondary variable, while porosity data of wells was used as a primary variable. In addition, a distribution of fractures in the secondary porosity model was also governed by geo-tectonic parameters such as dip, azimuth, major and minor values of fault systems or fractured zones [5].
Co-Kriging is a method that is used for simulation of oil-fi eld by interpolating algorithms, which based on the analysis of error changes with distances.
Co-Kriging solution is used to integrate the primary variable with secondary variables by calculating correlation coeffi cients and the experimental variogram statistics function. Figs.14 and 15 present a very good correlation between well fracture porosity (red curves) and the fracture porosity model of the Hai Su Den basement, which was built by integrating ANN and Co-Kriging techniques.
In order to assess the reliability of the porosity model, a series of correlations were carried out. The distribution and the relationship between well fracture porosity and porosity received from ANN and Co-Kriging show a very high degree of correlation (Fig.16).
5. Conclusions
The quality of Hai Su Den 3D 2009 Beam PSDM seismic data is Fig.15. Porosity model and depth slices using Co-Kriging technique
of Hai Su Den fractured basement
Fig.16. Comparison between secondary porosity modelling using Co-Kriging technique and well’s porosity on section
Fig.14. Vertical section of porosity model using Co-Kriging technique of Hai Su Den fractured basement
good for the basement; as a result many seismic attributes could have positive results in qualitative and quantitative analysis for identifi cation and outlining of good fracture zones inside the basement and for characterisation of the fracture systems.
The seismic attribute set which can optimise fracture porosity has been selected, consisting of the three following attributes: Refl ection Intensity, Gradient Magnitude, Sweetness.
In the case of good seismic data as in the Hai Su Den fi eld, integration of ANN and Co-Kriging techniques was proved as a good method to build fracture porosity model for the fractured basement reservoir.
References
1. J.N.Ogbechie. Fracture modelling and fl ow behaviour in shale gas reservoirs using Discrete Fracture Networks. 2011: p. 27 - 29.
2. Mai Thanh Ha. Seismic attribute analysis and its applications on data from the Canada, Mexico, USA, and Vietnam. University of Oklahoma. 2010: p. 30 - 32.
3. Nguyen Huy Ngoc, Sahalan B.Aziz, Nguyen Anh Duc. The application of seismic attributes for reservoir characterization in Pre-Tertiary fractured basement, Vietnam - Malaysia off shore. Interpretation. February 2014; 2(1).
4. Nguyen Huy Ngoc, Nguyen Quoc Quan, Hoang Ngoc Dong, Pham Huy Long, Tran Nhu Huy. Application
of “From seismic interpretation to tectonic reconstruction”
methodology to study Pre-Tertiary fractured granite basement reservoir in Cuu Long basin, Southeast Vietnam off shore. AAPG International Conference and Exhibition, Rio de Janeiro, Brazil. 15 - 18 November, 2009.
5. Nguyen Huy Ngoc, Nguyen Quoc Quan, Hoang Ngoc Dong, Nguyen Do Ngoc Nhi. Role of 3D seismic data in prediction of high potential areas within Pre-Tertiary fractured granite base-ment reservoir in Cuu Long basin, Vietnam off shore. AAPG International Conference and Exhibition, Calgary, Alberta, Canada. 12 - 15, September, 2010.
6. Nguyen Lam Anh. Integration of well and seismic data in building 3D geological model for fracture reservoir of White Tiger oil fi eld. The 2nd International Conference
“Fracture basement reservoir”, Vung Tau, Vietnam.
September, 2008.
7. T.Taner. Seismic attributes. CSEG Recoder. 2001:
p. 50.
8. Trịnh Xuân Cường, Hoàng Văn Quý. Mô hình hóa đá chứa móng nứt nẻ. Tạp chí Dầu khí. 2008; 5: trang 12 - 18.
9. W.J.Schmidt, Nguyễn Văn Quế, Phạm Huy Long.
Tiến hóa kiến tạo bể Cửu Long, Việt Nam. Tuyển tập Báo cáo Hội nghị Khoa học - Công nghệ: Viện Dầu khí Việt Nam 25 năm xây dựng và trưởng thành. 2003: trang 87 - 109.
